THIRST Alert Trial

Overview

The aim of this feasibility study is to determine whether an alert embedded within the electronic health record (EHR) causes clinicians to enrol patients into a randomised controlled trial (RCT) comparing oral fluid restriction versus no restriction in patients admitted to hospital with fluid overload. One of the main causes of fluid overload is heart failure where there is a lack of strong evidence to support the effectiveness of oral fluid restriction in the acute setting. This causes significant variation in clinical practice where decisions on whether or not to impose a restriction in oral fluid intake is based on the preference of the treating clinician rather than robust evidence from research. THIRST Alert is a pragmatic randomised controlled trial (RCT), embedded in the EHR, which seeks to determine whether a computerised alert for the clinical team can change clinician behaviour during routine NHS care at University College London Hospitals NHS Foundation Trust (UCLH). Patients with suspected fluid overload will be identified based on the prescription of intravenous furosemide, a medication used to stimulate diuresis (increased urine output) to remove excess fluid. A repeat prescription of intravenous furosemide within the first 48 hours of an unplanned admission will trigger the alert. A clinician from the treating team will then be asked to consider enrolling the patient into the RCT if they judge that oral fluid restriction might be beneficial but they have uncertainty about this (clinical equipoise). Enrolled patients will be randomised to either oral fluid restriction of 1 litre per day or no fluid restriction. This will then be actioned through documenting as part of the clinical plan in the patients record and then communicated to the patient and the rest of the clinical team, including nursing staff. The study will record the number of patients recruited into the trial and the effect of the alert on enrolled patients' subsequent oral fluid intake. There are no additional tests or follow up for patients and the trial finishes on discharge from the study site. All trial outcomes will use data collected from routine care and the study is supported by the UCLH Biomedical Research Centre, funded by NIHR.

Trial Design: A single centre, pragmatic, parallel group randomised controlled trial and feasibility study. The study will be conducted entirely within University College Hospital, at University College London Hospitals NHS Foundation Trust. Research Hypothesis: Clinician-facing alerts, delivered through the electronic health records system, provide a feasible method of enrolling patients into pragmatic trials evaluating the effectiveness of routine treatments Clinical Research question: In hospitalised patients with fluid overload, does an automated clinician-facing computerised decision support system (CDSS) alert recommendation for either oral fluid restriction (to 1L per day) or no-fluid restriction (continue with free fluids) result in a change in clinical behaviour and change in the oral fluid intake of enrolled patients? Study participants: The study population will be patients treated for suspected fluid overload with a regular prescription of intravenous (IV) furosemide within 48 hours of an unplanned hospital admission. For staff participants, any clinician with prescribing rights for IV furosemide may be exposed to the CDSS alert. For subsequent medical staff, nursing staff and other allied health professionals, patient enrolment into the trial and their treatment allocation status will be communicated through both EHR and clinical processes that are embedded within routine care, including verbal and non-digital patient handover. Summary of Interventions: This study will evaluate the effect of a Best practice advisory (BPA) interruptive alert that is embedded within the electronic health record (EHR) used at the study site. The first BPA will ask the treating clinician whether they judge randomisation to either oral fluid restriction or no restriction is appropriate - if yes is selected, randomised allocation to one of two subsequent alerts will occur at the point of care. This intervention will be tested in silico prior to deployment to the live EHR. Ethics: This study protocol was approved by the London Riverside Research Ethics Committee (Ref: 22/LO/0889) and sponsored by University College London (Ref: 151938).